implemented SRT Parser
This commit is contained in:
Luis Miguel Almánzar
parent
4e710ec0a4
commit
f2237b858a
|
|
@ -1,17 +1,61 @@
|
||||||
using System;
|
using System;
|
||||||
using System.Collections.Generic;
|
using System.Collections.Generic;
|
||||||
|
using System.Globalization;
|
||||||
using System.IO;
|
using System.IO;
|
||||||
using System.Linq;
|
using System.Text.RegularExpressions;
|
||||||
using System.Text;
|
|
||||||
using System.Threading.Tasks;
|
|
||||||
|
|
||||||
namespace MediaBrowser.MediaEncoding.Subtitles
|
namespace MediaBrowser.MediaEncoding.Subtitles
|
||||||
{
|
{
|
||||||
public class SrtParser : ISubtitleParser
|
public class SrtParser : ISubtitleParser
|
||||||
{
|
{
|
||||||
public SubtitleTrackInfo Parse(Stream stream)
|
private readonly CultureInfo _usCulture = new CultureInfo("en-US");
|
||||||
{
|
public SubtitleTrackInfo Parse(Stream stream) {
|
||||||
throw new NotImplementedException();
|
var trackInfo = new SubtitleTrackInfo();
|
||||||
|
using ( var reader = new StreamReader(stream))
|
||||||
|
{
|
||||||
|
string line;
|
||||||
|
while ((line = reader.ReadLine()) != null)
|
||||||
|
{
|
||||||
|
if (string.IsNullOrWhiteSpace(line))
|
||||||
|
{
|
||||||
|
continue;
|
||||||
|
}
|
||||||
|
var subEvent = new SubtitleTrackEvent {Id = line};
|
||||||
|
line = reader.ReadLine();
|
||||||
|
var time = Regex.Split(line, @"[\t ]*-->[\t ]*");
|
||||||
|
subEvent.StartPositionTicks = GetTicks(time[0]);
|
||||||
|
var endTime = time[1];
|
||||||
|
var idx = endTime.IndexOf(" ", StringComparison.Ordinal);
|
||||||
|
if (idx > 0)
|
||||||
|
endTime = endTime.Substring(0, idx);
|
||||||
|
subEvent.EndPositionTicks = GetTicks(endTime);
|
||||||
|
var multiline = new List<string>();
|
||||||
|
while ((line = reader.ReadLine()) != null)
|
||||||
|
{
|
||||||
|
if (string.IsNullOrEmpty(line))
|
||||||
|
{
|
||||||
|
break;
|
||||||
|
}
|
||||||
|
multiline.Add(line);
|
||||||
|
}
|
||||||
|
subEvent.Text = string.Join(@"\N", multiline);
|
||||||
|
subEvent.Text = Regex.Replace(subEvent.Text, "\\{(\\\\[\\w]+\\(?([\\w\\d]+,?)+\\)?)+\\}", string.Empty, RegexOptions.IgnoreCase);
|
||||||
|
subEvent.Text = Regex.Replace(subEvent.Text, "<", "<", RegexOptions.IgnoreCase);
|
||||||
|
subEvent.Text = Regex.Replace(subEvent.Text, ">", ">", RegexOptions.IgnoreCase);
|
||||||
|
subEvent.Text = Regex.Replace(subEvent.Text, "<(\\/?(font|b|u|i|s))((\\s+(\\w|\\w[\\w\\-]*\\w)(\\s*=\\s*(?:\\\".*?\\\"|'.*?'|[^'\\\">\\s]+))?)+\\s*|\\s*)(\\/?)>", "<$1$3$7>", RegexOptions.IgnoreCase);
|
||||||
|
subEvent.Text = Regex.Replace(subEvent.Text, @"\\N", "<br />",RegexOptions.IgnoreCase);
|
||||||
|
trackInfo.TrackEvents.Add(subEvent);
|
||||||
|
}
|
||||||
|
}
|
||||||
|
return trackInfo;
|
||||||
|
}
|
||||||
|
|
||||||
|
long GetTicks(string time) {
|
||||||
|
TimeSpan span;
|
||||||
|
return TimeSpan.TryParseExact(time, @"hh\:mm\:ss\.fff", _usCulture, out span)
|
||||||
|
? span.Ticks
|
||||||
|
: (TimeSpan.TryParseExact(time, @"hh\:mm\:ss\,fff", _usCulture, out span)
|
||||||
|
? span.Ticks : 0);
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
|
||||||
|
|
@ -50,6 +50,7 @@
|
||||||
</Otherwise>
|
</Otherwise>
|
||||||
</Choose>
|
</Choose>
|
||||||
<ItemGroup>
|
<ItemGroup>
|
||||||
|
<Compile Include="MediaEncoding\Subtitles\SrtParserTests.cs" />
|
||||||
<Compile Include="Providers\MovieDbProviderTests.cs" />
|
<Compile Include="Providers\MovieDbProviderTests.cs" />
|
||||||
<Compile Include="Resolvers\MovieResolverTests.cs" />
|
<Compile Include="Resolvers\MovieResolverTests.cs" />
|
||||||
<Compile Include="Resolvers\TvUtilTests.cs" />
|
<Compile Include="Resolvers\TvUtilTests.cs" />
|
||||||
|
|
@ -61,6 +62,10 @@
|
||||||
<Project>{17e1f4e6-8abd-4fe5-9ecf-43d4b6087ba2}</Project>
|
<Project>{17e1f4e6-8abd-4fe5-9ecf-43d4b6087ba2}</Project>
|
||||||
<Name>MediaBrowser.Controller</Name>
|
<Name>MediaBrowser.Controller</Name>
|
||||||
</ProjectReference>
|
</ProjectReference>
|
||||||
|
<ProjectReference Include="..\MediaBrowser.MediaEncoding\MediaBrowser.MediaEncoding.csproj">
|
||||||
|
<Project>{0BD82FA6-EB8A-4452-8AF5-74F9C3849451}</Project>
|
||||||
|
<Name>MediaBrowser.MediaEncoding</Name>
|
||||||
|
</ProjectReference>
|
||||||
<ProjectReference Include="..\MediaBrowser.Model\MediaBrowser.Model.csproj">
|
<ProjectReference Include="..\MediaBrowser.Model\MediaBrowser.Model.csproj">
|
||||||
<Project>{7eeeb4bb-f3e8-48fc-b4c5-70f0fff8329b}</Project>
|
<Project>{7eeeb4bb-f3e8-48fc-b4c5-70f0fff8329b}</Project>
|
||||||
<Name>MediaBrowser.Model</Name>
|
<Name>MediaBrowser.Model</Name>
|
||||||
|
|
@ -77,6 +82,11 @@
|
||||||
<ItemGroup>
|
<ItemGroup>
|
||||||
<None Include="app.config" />
|
<None Include="app.config" />
|
||||||
</ItemGroup>
|
</ItemGroup>
|
||||||
|
<ItemGroup>
|
||||||
|
<None Include="MediaEncoding\Subtitles\TestSubtitles\unit.srt">
|
||||||
|
<CopyToOutputDirectory>Always</CopyToOutputDirectory>
|
||||||
|
</None>
|
||||||
|
</ItemGroup>
|
||||||
<Choose>
|
<Choose>
|
||||||
<When Condition="'$(VisualStudioVersion)' == '10.0' And '$(IsCodedUITest)' == 'True'">
|
<When Condition="'$(VisualStudioVersion)' == '10.0' And '$(IsCodedUITest)' == 'True'">
|
||||||
<ItemGroup>
|
<ItemGroup>
|
||||||
|
|
|
||||||
108
MediaBrowser.Tests/MediaEncoding/Subtitles/SrtParserTests.cs
Normal file
108
MediaBrowser.Tests/MediaEncoding/Subtitles/SrtParserTests.cs
Normal file
|
|
@ -0,0 +1,108 @@
|
||||||
|
using System;
|
||||||
|
using System.Collections.Generic;
|
||||||
|
using System.IO;
|
||||||
|
using MediaBrowser.MediaEncoding.Subtitles;
|
||||||
|
using Microsoft.VisualStudio.TestTools.UnitTesting;
|
||||||
|
|
||||||
|
namespace MediaBrowser.Tests.MediaEncoding.Subtitles {
|
||||||
|
|
||||||
|
[TestClass]
|
||||||
|
public class SrtParserTests {
|
||||||
|
|
||||||
|
[TestMethod]
|
||||||
|
public void TestParse() {
|
||||||
|
|
||||||
|
var expectedSubs =
|
||||||
|
new SubtitleTrackInfo {
|
||||||
|
TrackEvents = new List<SubtitleTrackEvent> {
|
||||||
|
new SubtitleTrackEvent {
|
||||||
|
Id = "1",
|
||||||
|
StartPositionTicks = 24000000,
|
||||||
|
EndPositionTicks = 52000000,
|
||||||
|
Text =
|
||||||
|
"[Background Music Playing]"
|
||||||
|
},
|
||||||
|
new SubtitleTrackEvent {
|
||||||
|
Id = "2",
|
||||||
|
StartPositionTicks = 157120000,
|
||||||
|
EndPositionTicks = 173990000,
|
||||||
|
Text =
|
||||||
|
"Oh my god, Watch out!<br />It's coming!!"
|
||||||
|
},
|
||||||
|
new SubtitleTrackEvent {
|
||||||
|
Id = "3",
|
||||||
|
StartPositionTicks = 257120000,
|
||||||
|
EndPositionTicks = 303990000,
|
||||||
|
Text = "[Bird noises]"
|
||||||
|
},
|
||||||
|
new SubtitleTrackEvent {
|
||||||
|
Id = "4",
|
||||||
|
StartPositionTicks = 310000000,
|
||||||
|
EndPositionTicks = 319990000,
|
||||||
|
Text =
|
||||||
|
"This text is <font color=\"red\">RED</font> and has not been positioned."
|
||||||
|
},
|
||||||
|
new SubtitleTrackEvent {
|
||||||
|
Id = "5",
|
||||||
|
StartPositionTicks = 320000000,
|
||||||
|
EndPositionTicks = 329990000,
|
||||||
|
Text =
|
||||||
|
"This is a<br />new line, as is<br />this"
|
||||||
|
},
|
||||||
|
new SubtitleTrackEvent {
|
||||||
|
Id = "6",
|
||||||
|
StartPositionTicks = 330000000,
|
||||||
|
EndPositionTicks = 339990000,
|
||||||
|
Text =
|
||||||
|
"This contains nested <b>bold, <i>italic, <u>underline</u> and <s>strike-through</s></u></i></b> HTML tags"
|
||||||
|
},
|
||||||
|
new SubtitleTrackEvent {
|
||||||
|
Id = "7",
|
||||||
|
StartPositionTicks = 340000000,
|
||||||
|
EndPositionTicks = 349990000,
|
||||||
|
Text =
|
||||||
|
"Unclosed but <b>supported HTML tags are left in, SSA italics aren't"
|
||||||
|
},
|
||||||
|
new SubtitleTrackEvent {
|
||||||
|
Id = "8",
|
||||||
|
StartPositionTicks = 350000000,
|
||||||
|
EndPositionTicks = 359990000,
|
||||||
|
Text =
|
||||||
|
"<ggg>Unsupported</ggg> HTML tags are escaped and left in, even if <hhh>not closed."
|
||||||
|
},
|
||||||
|
new SubtitleTrackEvent {
|
||||||
|
Id = "9",
|
||||||
|
StartPositionTicks = 360000000,
|
||||||
|
EndPositionTicks = 369990000,
|
||||||
|
Text =
|
||||||
|
"Multiple SSA tags are stripped"
|
||||||
|
},
|
||||||
|
new SubtitleTrackEvent {
|
||||||
|
Id = "10",
|
||||||
|
StartPositionTicks = 370000000,
|
||||||
|
EndPositionTicks = 379990000,
|
||||||
|
Text =
|
||||||
|
"Greater than (<) and less than (>) are shown"
|
||||||
|
}
|
||||||
|
}
|
||||||
|
};
|
||||||
|
|
||||||
|
var sut = new SrtParser();
|
||||||
|
|
||||||
|
var stream = File.OpenRead(@"MediaEncoding\Subtitles\TestSubtitles\unit.srt");
|
||||||
|
|
||||||
|
var result = sut.Parse(stream);
|
||||||
|
|
||||||
|
Assert.IsNotNull(result);
|
||||||
|
Assert.AreEqual(expectedSubs.TrackEvents.Count,result.TrackEvents.Count);
|
||||||
|
for (int i = 0; i < expectedSubs.TrackEvents.Count; i++)
|
||||||
|
{
|
||||||
|
Assert.AreEqual(expectedSubs.TrackEvents[i].Id, result.TrackEvents[i].Id);
|
||||||
|
Assert.AreEqual(expectedSubs.TrackEvents[i].StartPositionTicks, result.TrackEvents[i].StartPositionTicks);
|
||||||
|
Assert.AreEqual(expectedSubs.TrackEvents[i].EndPositionTicks, result.TrackEvents[i].EndPositionTicks);
|
||||||
|
Assert.AreEqual(expectedSubs.TrackEvents[i].Text, result.TrackEvents[i].Text);
|
||||||
|
}
|
||||||
|
|
||||||
|
}
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
@ -0,0 +1,44 @@
|
||||||
|
|
||||||
|
|
||||||
|
1
|
||||||
|
00:00:02.400 --> 00:00:05.200
|
||||||
|
[Background Music Playing]
|
||||||
|
|
||||||
|
2
|
||||||
|
00:00:15,712 --> 00:00:17,399 X1:000 X2:000 Y1:050 Y2:100
|
||||||
|
Oh my god, Watch out!
|
||||||
|
It's coming!!
|
||||||
|
|
||||||
|
3
|
||||||
|
00:00:25,712 --> 00:00:30,399
|
||||||
|
[Bird noises]
|
||||||
|
|
||||||
|
4
|
||||||
|
00:00:31,000 --> 00:00:31,999
|
||||||
|
This text is <font color="red">RED</font> and has not been {\pos(142,120)}positioned.
|
||||||
|
|
||||||
|
5
|
||||||
|
00:00:32,000 --> 00:00:32,999
|
||||||
|
This is a\nnew line, as is\Nthis
|
||||||
|
|
||||||
|
6
|
||||||
|
00:00:33,000 --> 00:00:33,999
|
||||||
|
This contains nested <b>bold, <i>italic, <u>underline</u> and <s>strike-through</s></u></i></b> HTML tags
|
||||||
|
|
||||||
|
7
|
||||||
|
00:00:34,000 --> 00:00:34,999
|
||||||
|
Unclosed but <b>supported HTML tags are left in, {\i1} SSA italics aren't
|
||||||
|
|
||||||
|
8
|
||||||
|
00:00:35,000 --> 00:00:35,999
|
||||||
|
<ggg>Unsupported</ggg> HTML tags are escaped and left in, even if <hhh>not closed.
|
||||||
|
|
||||||
|
9
|
||||||
|
00:00:36,000 --> 00:00:36,999
|
||||||
|
Multiple {\pos(142,120)\b1}SSA tags are stripped
|
||||||
|
|
||||||
|
10
|
||||||
|
00:00:37,000 --> 00:00:37,999
|
||||||
|
Greater than (<) and less than (>) are shown
|
||||||
|
|
||||||
|
|
||||||
Loading…
Reference in New Issue
Block a user